Magnetoencephalography

Magnetoencephalography

Malcolm Proudfoot,1,2 Mark W Woolrich,2 Anna C Nobre,2

Martin R Turner1

1Nuffield Department of ClinicalNeurosciences, University ofOxford, UK2Oxford Centre for Human BrainActivity, University of Oxford, UK

Correspondence toDr Martin Turner, NuffieldDepartment of ClinicalNeurosciences, West WingLevel 3, John Radcliffe Hospital,Oxford OX3 9DU, UK;[email protected]

To cite: Proudfoot M,Woolrich MW, Nobre AC,et al. Pract Neurol PublishedOnline First: [please includeDay Month Year]doi:10.1136/practneurol-2013-000768

INTRODUCTIONThe understanding of brain function ismoving rapidly towards a systems-level,network-based approach. It is now as naiveto talk simplistically about what a particu-lar area of the brain ‘does’, as it was forFranz Joseph Gall (1758–1828) to linkits performance to the thickness of theoverlying skull. Magnetoencephalography(MEG) is a rapidly developing and uniquetool for the study of brain function, inparticular the underlying oscillations inneuronal activity that appear to be funda-mental (box 1), with real-time resolutionand potential for application across a rangeof brain disorders. We provide a briefoverview of the technology, broadapproaches to data analysis, and aspirationsfor its application to the study ofneurodegeneration.

FUNCTIONAL BRAIN IMAGING SO FARStructural MR imaging of the brain andspinal cord has revolutionised the accur-acy of diagnosis in common conditions,such as stroke, and greatly expanded thetaxonomy of neurological disorders.Advanced applications of MRI now allowthe assessment of white matter tractintegrity (diffusion-tensor imaging),regional grey matter volume (voxel-basedmorphometry), and cortical thickness(surface-based morphometry). Thesetechniques enable the non-invasive andrapid quantification of structure at agiven time point, but it is clear that thebrain cannot be understood in terms ofstructure alone. Functional MRI (fMRI),based on blood oxygen-level-dependent(BOLD) image contrast, can achievealmost submillimetre accuracy in thespatial localisation of neuronal activity.However, this relatively high spatial reso-lution is not matched in temporal accur-acy, in essence because the relatively slowspeed of haemodynamic changes inresponse to neuronal activity fundamen-tally limits the temporal detail that can beextracted to a timescale of seconds.When one considers the multiple synaptic

transmissions and physical distancecovered by brain activity within this timeframe, it is quickly appreciated that thistechnology is limited in its ability todeliver a systems-level understanding ofbrain function if used in isolation.

THE UNIQUE ADVANTAGES OF MEGEver since the pioneering EEG recordingsmade by German neurologist HansBerger (1873–1941), it has been possibleto identify the self-generated oscillatoryactivity of neuronal ensembles and cat-egorise frequency bands with increasingaccuracy. Electrical potential changesrelated to brain activity measured at thescalp by EEG are fundamentally limited

Box 1 An introduction to neuronaloscillations

▸ Neuronal oscillatory activity is continu-ous, but fluctuations in power andtiming allow rapid alteration in com-munication strength within existingstructural network architecture, farfaster than synaptic modification.1

▸ Two distinct cerebral regions can facili-tate preferential information exchangeby synchronising their rhythmic behav-iour; the γ band (40–80 Hz), in particu-lar, facilitates this process, but is alsomodulated ‘top-down’ by lower fre-quencies such as θ (4–7 Hz), reflectingfactors, such as arousal states.

▸ α Rhythms (8–13 Hz), so prominent inthe occipital cortex upon eye closure,reflect more than just an ‘idling’ rhythmbut also contribute to active allocationof attentional resources and suppressirrelevant sensory information.2

▸ The influential theory ‘Communicationthrough Coherence’ developed byFries,3 builds on existing models of‘binding by synchronisation’ that mayunderpin selective attention, a key func-tion in prioritising neural events toguide awareness and action.4

HOW TO UNDERSTAND IT

Proudfoot M, et al. Pract Neurol 2014;0:1–8. doi:10.1136/practneurol-2013-000768 1

by the distortive effects of the intervening structures,which severely hamper efforts to localise the signalsource precisely. MEG, instead, measures the magneticfield changes induced by intracellular current flow, thegeneration of which obeys the ‘right-hand rule’ in theapplication of Ampère’s law. Unlike EEG measures,these pass through dura, skull and scalp relativelyunaltered. The technique, therefore, offers a safe,non-invasive method to ‘listen’ in to brain activity atrest and during simple tasks, which from the subject’sperspective, despite measuring at several hundredchannels, is painless and quick to set up.Mathematical modelling of these data then enableslocalisation of sources while uniquely maintainingsampling frequencies up to several thousand times persecond. Compared to fMRI’s temporal resolution of,at best, several hundred milliseconds, MEG canresolve events with millisecond precision.

HOW IT WORKSThe neuronal activity captured by MEG is not, asperhaps expected, generated by the (too brief ) axonalaction potentials of pyramidal cells, but rather by thenet contributions of excitatory and inhibitory den-dritic postsynaptic potentials. This current flowthrough the apical dendrites (represented as a‘dipole’) generates a magnetic field that projects radi-ally; thus, MEG excels at detecting dipoles arrangedin a tangential orientation to the skull. Fortunately,the extensively folded sulci of the human cortexpromote that orientation for the majority of corticalmicrocolumns (figure 1). However, MEG is less sensi-tive to deeper (including subcortical) sources, as mag-netic field change decreases rapidly with distance.Compared with a standard clinical MR scanner

magnet strength of 1.5 Tesla, the strength of thesignals detected by MEG are 1014 orders smaller(figure 2). It has been compared with hearing a pindrop at a rock concert. The smallest measurable mag-netic field changes are thought to be produced by sim-ultaneously active arrays of approximately 50 000pyramidal cells, which in theory covers a corticalsurface area of 0.9 mm diameter. It is increasinglyrecognised that modulation of self-generated oscilla-tory activity is a principal mechanism by which geo-graphically distant network regions interact,6 thus, abrain-wide imaging technique with high temporal sen-sitivity is a prerequisite for interrogation.The ability to detect endogenously generated mag-

netic fields was realised in the 1960s by physicistDavid Cohen at Massachusetts Institute ofTechnology, who furthered the then recent discoveryof ‘magnetocardiography’ by applying a magneticallyshielded room to remove the overwhelming noise ofthe Earth’s magnetic field (figure 3). He could thenmeasure the even smaller magnetoencephalographicsignal by making use of superconducting loops super-conducting quantum interference device (SQUIDs),

developed by his collaborator James Zimmerman. Atvery low temperatures, SQUIDs are extremely sensi-tive to magnetic field change, which can be recordedand converted into digital signal (‘quantisation’).Sensor arrays have evolved to provide whole-headcoverage via a helmet containing more than 300

Figure 2 Magnetic field strength density measured infemtotesla (fT), highlighting the exquisite sensitivity of theSQUIDs used in magnetoencephalography.

Figure 1 The organisation of cortical microcolumns within thesulcal bank, tangentially orientated to the skull, allows theirdetection with magnetoencephalography since their inducedmagnetic fields will project beyond the skull surface. Conversely,apical dendrites orientated perpendicularly to the skull, asfound at the gyral crown, are better detected by EEG. (FromHansen et al5).


2 Proudfoot M, et al. Pract Neurol 2014;0:1–8. doi:10.1136/practneurol-2013-000768

sensor sites, enveloped in a ‘dewar’ of cooling liquidhelium. The subject’s head is positioned underneaththe helmet (figure 4) and it is possible to acquire EEGrecordings simultaneously if desired. In the future,MEG may include atomic sensors, which still exploitthe principle of quantum tunnelling of pairs of elec-trons, but obviate the need for cryogenictemperatures.7

THE ACQUISITIONThe sensor design itself has evolved to meet some ofthe localisation challenges by using more than onepick-up coil in series. A single ‘magnetometer’ coilmeasures any orthogonal magnetic field. Pairs of coilsplace closed together and wound in opposite direc-tions are also used to measure gradients in the mag-netic field over space. These ‘gradiometers’ areparticularly sensitive to a gradient in magnetism fromnearby (ie, neuronal) sources, but subtract out signalfrom distant external (and thus artefactual) sources, asthese appear similar to both coils.Electromagnetic signals are also generated by move-

ment of the head or eyes (including blinking), skeletaland cardiac muscle electromagnetic activity. Thesensors, therefore, also pick up this physiologicalnoise, unrelated to brain activity. Interference fromdental amalgam, metal zippers, jewellery, and bra fas-teners is also significant and avoided by pre-scanscreens if possible. Head location must be calculatedrelative to the SQUID sensors, since the helmet is nota tight fit. This is achieved by using small magneticcoils attached to anatomical landmarks to localise thesubject’s head position in the MEG scanner, and thesecan also be used to enable continuous motion correc-tion. However, it is also important to maximisesubject comfort to discourage excess head movement.Considerable care must also be taken in ensuring

that all stimulus presentation and response deviceswithin the magnetically shielded room are themselveselectromagnetically silent. Experimental tasks are typ-ically designed to avoid eye movements during rele-vant measurement periods in order to minimiseartefact, but remaining eye movements and blinks canstill be identified with a combination of surfaceelectro-oculography and infrared eye tracking. AnECG is typically also recorded to enable identificationand subsequent exclusion of cardiac electrical activity,which can otherwise create large contaminating arte-facts in the MEG signal. There are comprehensiveexpert consensus guidelines aiming to harmoniseMEG experimental strategies across sites8 in an effortto improve reproducibility.

THE ANALYSISMEG analysis involves enormous datasets that requirevast computer processing power to manipulate them.Having overcome the initial difficulties in artefactidentification, MEG data analysis still involves consid-erable complexity. The fundamental issue is that ofthe ‘inverse problem’. This concept, in relation toMEG, summarises the challenge of precisely localisingin three dimensional (3D) space the underlying neuralsources of a magnetic recording. In reality, there maybe many equally plausible combinations of neuralsources, and thus, without additional constraints thesolution is not unique. Yet, this is a challenge our ownbrains overcome daily by using constraints that reflect

Figure 3 After 4 s of raw magnetoencephalography data (twochannels contain obvious artefacts), the door to themagnetically shielded room is opened during recording. Theinterference caused by external magnetic fields highlights whyeffective room shielding is essential.

Figure 4 A MEG recording session at OHBA, eyetracker deviceshown in foreground.



sensible prior assumptions, for example in deciding ifa visual object is small and close, or large and faraway. We have existing expectations about object sizeto guide us.Methods to solve the inverse problem, therefore,

need to make additional assumptions, such as thebrain activity being spatially sparse or smooth. Theappropriate assumptions to make are often directedby the particular experimental protocol and expectedfindings. A given paradigm may seek to identify differ-ential brain activity during a rudimentary sensorystimulation task, in which case the localisation modelmight assume that a small number of dipole sourcescould account for the difference in signal production(ie, the brain activity is sparse). Alternatively, forexample during cognitive tasks, there is a wide net ofpossible sources in the brain to consider, and we needmore advanced modelling techniques. This includesapproaches that assume the brain activity is smoothand sparse,9 and also constrained to a 3D map of thepatient’s own cortical mantle (obtained during a sub-sequent structural MRI). A common approach forrepresenting these various assumptions is to use priorsin powerful Bayesian algorithms. A popular alternativeto overcome the inverse problem is to use beamform-ing. This method, originally developed for use inradar arrays, corresponds to an adaptive spatial filterdesigned to extract the origins of a signal from someprespecified spatial location.10

MEG data in either ‘sensor space’ (presented asdata recorded across the distribution of the sensors)or ‘source space’ (reconstructed to a 3D model ofbrain sources) offer a wealth of analysis possibilitiesand selected features that can relate to a particularclinical question (figure 5).For example, it is possible to analyse the data time-

locked to experimental events in terms of event-relatedfields (ERF, the MEG analogue of event-related poten-tials), such as the mismatch negativity abnormalitiesdescribed in dyslexia.11 Additionally, MEG is

particularly powerful at investigating oscillatory brainactivity by using ‘Fourier transformations’, or relatedtransforms, such as Wavelet or Hilbert, to describe thesignal in terms of how it oscillates in different fre-quency bands over time. These frequency bands showmeaningful changes in power or synchronisation ofongoing oscillatory activity at behaviourally relevanttime points, and can also form the basis of functionalconnectivity measurements (correlated changes overtime) between different brain regions captured inMEG data12 (figure 6).

THE APPLICATIONS AND POTENTIAL OF MEGMEG has found early clinical application in epilepticsource localisation, in particular for the planning ofsubsequent surgery, and shows close correlation toinvasive studies of cortical activity.13 In a new era,MEG applications are no longer limited to descriptionof relatively fundamental neurophysiology, but haveinstead begun to describe whole-brain activity at thenetwork-level during the so-called resting state14

(figure 7) as well as task performance.15

This approach has been pioneered by fMRI,16 butMEG now validates these findings in eliminating arte-factual explanations based solely on correlated patternsof vascular activity. MEG also offers the possibility ofdecomposing the time-course of communicationbetween ‘nodes’ in such networks, assessing their func-tionality during task activity, and indeed describingtheir dysfunction during neurodegenerative disease,17

with the ultimate hope of developing sensitive pharma-codynamic markers of therapeutic response. MEGalso offers information complementary to fMRI; theuncertain impact of disease on the time course of theBOLD haemodynamic response function encouragesmultimodal data acquisition during brain activation (by

Figure 5 Sensor space MEG data (A) presented as a twodimensional (2D) topographic map of contrasted α (8–12 Hz)activity in a visual attention task (data concatenated from 38subjects). α power is lower in the contralateral (attending)hemisphere. MEG data reconstructed into source space (B) witha beamformer approach and presented on a 3D cortical map.

Figure 6 Time–frequency plot from posterior sensorsfollowing presentation of a visual stimulus (multiple subjectscombined). Increased power is noted in θ followed shortly by γ.Desynchronisation instead occurs in α and β bands. Vertical linesdenote stimulus onset/offset.



necessity non-concurrent, as MEG and fMRI cannotbe acquired simultaneously).18

MEG correlates of neurodegenerationFocal slowing in temporoparietal regions was anexpected finding in early MEG studies of Alzheimer’sdisease, and was furthermore correlated with cogni-tive measures.19 Attempts to model the source ofspontaneous α band activity also noted an anteriorshift from parieto-occipital regions to predominantlytemporal-lobe generators in Alzheimer’s disease versuscontrols; this was interpreted to reflect somehow aloss of cholinergic transmission.20 Another study pro-vided more convincing evidence that described lefttemporal MEG activity deficits, which correlated withipsilateral hippocampal atrophy and behaviouralmeasures.21

Surface EEG had already described abnormalities inpreattentive auditory processing of deviant tones inneurodegenerative disease. These excessive evoked

response potential (ERP) abnormalities, perhapsreflecting a failure to inhibit or adapt to irrelevantstimuli, were convincingly replicated and localised inMEG studies of Parkinson’s disease,22 Alzheimer’sdisease23 and amyotrophic lateral sclerosis.24

Subjects with Parkinson’s disease, even withoutdementia, also show widespread oscillatory slowing.25

Furthermore, in combination with invasive localrecordings, MEG identified a dopamine-responsivelong-distance network operating in the β band thatshowed coherence between prefrontal cortex and sub-thalamic nuclei, in contrast to a spatially and spec-trally distinct network in the α band connectingbrainstem with temporoparietal cortex.26

Global resting-state networks in neurodegenerationA growing number of MEG studies have describedneurodegenerative alteration to functional connectiv-ity during resting-state recordings, complementingfindings from fMRI. Resting-state networks describe

Figure 7 Comparison of resting-state networks identified from magnetoencephalography and fMRI data independently. (A) Defaultmode network, (B) left lateral frontoparietal network, (C) right lateral frontoparietal network, (D) sensorimotor network. (Adaptedfrom Brookes et al14).



the temporally coordinated, but often anatomicallydisparate, spontaneous fluctuations in brain activitythat are present even at rest within distinct functionalbrain networks.27 Although such networks are obvi-ously engaged during relevant task activity, study ofresting-state networks is particularly appealing withinpatient populations as impaired task performance isno longer a possible confounding factor.Alzheimer’s disease has been most frequently scruti-

nised using resting-state measures. A study of 18patients (mean Mini-Mental State Examination Score19.2) noted a loss of long-distance interhemisphericconnectivity (measured as synchronisation likelihood)within the α and β bands. This correlated with cogni-tive impairment and was not consequent to anticholi-nesterase treatment.28 A repeat study focused on themodular organisation of resting-state networks(decomposed into functional subnetworks), and notedAlzheimer’s disease to cause decrement of intermodu-lar connectivity but also intramodular damagerestricted to certain cortical regions.29

Such graph-theory concepts could, in principle, beemployed to transfer descriptions of a ‘connectome’(our unique neural fingerprint that defines us as indivi-duals) across different modalities, including fMRI anddiffusion tensor imaging (DTI) studies.30 However,results from these connectivity studies often conflict,perhaps reflecting differing population samples, but alsoimplying a mismatch between functional and structuralconnection strength,31 which may hint at underlyingpathological mechanisms, such as interneuronal dys-function in amyotrophic lateral sclerosis.32

Oscillatory signatures in neurodegenerationThe suitability of MEG to capture neuronal activityduring relevant motor or cognitive tasks has encour-aged parallel investigation of induced changes in oscil-latory activity—perturbations of spontaneous brainactivity patterns are fundamental to all our experi-ences, decisions and actions.1 A block-design study ofpatients with Alzheimer’s disease or dementia withLewy bodies contrasted their cortical spectral powerduring periods of either rest or repeated auditoryattention tasks. In both conditions, there were largegroup differences from anterior sensors in a 3–7 Hz θband.33 An event-related experimental design wasused to study visual working memory tasks attemptedby Alzheimer’s disease and vascular dementiapatients.34 Those dementia subjects still able to com-plete the task successfully demonstrated (against con-trols) significantly delayed and stronger amplitude αdesynchronisation during the task epochs (albeit withlimited topographical localisation), in keeping withthe comparatively higher burden of cognitive process.Task-based MEG recordings of 12 subjects with fron-

totemporal dementia were compared with controls,while making categorical semantic judgements aboutvisually presented objects. There was an early

difference in temporoparietal cortex activity followedby a later reduction in frontoparietal activity in eachtask epoch, interpreted as corresponding to semanticinformation processing and subsequent action selec-tion.35 This result highlights the high temporal reso-lution achievable in MEG recordings such thatcomponent parts of a rapid cognitive task can be distin-guished with confidence in time and anatomical space.Even swallowing has proved to be an informative

motor task in the study of neurodegenerative diseasewith MEG. An investigation of Parkinson’s disease sub-jects with and without dysphagia found evidence forcompensatory adaptive cerebral changes to involvewider cortical regions.36 The same researchers havenoted right hemispheric lateralisation of swallow-relatedactivity in amyotrophic lateral sclerosis37 and Kennedy’ssyndrome,38 perhaps reflecting plasticity in the face ofprogressive neurodegeneration.

Biomarker potentialMEG has been used to predict whether subjects weremore likely to progress to clinically definedAlzheimer’s disease on the basis of more widespreadpower changes during a memory task, in a study of asmall number of subjects with minimal cognitiveimpairment.39 A larger (117 Alzheimer’s disease sub-jects) multicentre MEG study was subsequently per-formed; 1 min of resting-state data was sufficient todetect increased functional connectivity.40 Ten-monthinterval assessment of 31 subjects detailed progressiveabnormalities, correlating with worsening neuropsy-chometry. A longitudinal study of Parkinson’s diseaseshowed progressive slowing of oscillatory activity overa 4-year follow-up period in 59 initially non-demented Parkinson’s disease subjects.41 Thesechanges were particularly associated with mild cogni-tive decline and, furthermore, MEG signals, aboveand beyond neuropsychometry, predicted subsequentconversion to Parkinson’s disease dementia, as borneout at the 7-year follow-up time point.42

CONCLUSIONIn combination with the increasingly high structuralresolution offered by MRI, MEG has unique potentialas part of a multimodal approach to brain disordersthat is sensitive to time as well as space. It seems clearfrom presymptomatic studies across a range of neuro-degenerative disorders, that the earliest events occurmany years, possibly decades, before the onset ofsymptoms. Primary prevention of neurodegenerationwill require sensitivity to very subtle changes in brainactivity that seem likely to operate at the networklevel.43 As well as studying the patterns of brain activ-ity in the resting state, cognitive or motor MEG-basedtasks could reveal key pathological or compensatorypatterns of activity that might form the basis for earlyintervention, including pharmacodynamic biomarkersof therapeutic intervention.



Key points

▸ Magnetoencephalography (MEG) is non-invasive, safeand comfortable.

▸ MEG offers unsurpassed temporal resolution; it canprobe neuronal oscillation activity that is fundamen-tal to brain function in health and disease.

▸ Modern MEG systems include several hundreddistortion-free sensors surrounding the head toprovide high spatial precision over superficial corticalareas.

▸ MEG analysis describes in vivo function of wholebrain networks in real time, and has the potential todetect the very earliest changes in neurodegenerativedisorders and assess preventative therapies of thefuture.

Acknowledgements We thank George Wallis, OHBA, forproviding figures 5 and 6.

Contributors MP drafted the manuscript. MWWedited themanuscript. KN edited the manuscript. MRT conceived andedited the manuscript, and is guarantor of the content.

Funding This work was support by the Wellcome Trust(092753), the National Institute for Health Research (NIHR)Oxford Biomedical Research Centre (BRC) based at OxfordUniversity Hospitals and by an MRC UK MEG PartnershipGrant, MR/K005464/1. MP receives funding from TheGuarantors of Brain and the Oxford BRC. MRTreceivesfunding from the Medical Research Council and MotorNeurone Disease Association UK Lady Edith WolfsonFellowship (MR/K01014X/1).

Competing interests None.

Provenance and peer review Not commissioned; externallypeer reviewed. This paper was reviewed by Khalid Hamandi,Cardiff, UK.

Open Access This is an Open Access article distributed inaccordance with the terms of the Creative CommonsAttribution (CC BY 3.0) license, which permits others todistribute, remix, adapt and build upon this work, forcommercial use, provided the original work is properly cited.See: http://creativecommons.org/licenses/by/3.0/

REFERENCES1 Buzsaki G. Rhythms of the Brain. USA: OUP, 2011.2 Klimesch W. Α-band oscillations, attention, and controlled

access to stored information. Trends Cogn Sci 2012;16:606–17.3 Fries P. A mechanism for cognitive dynamics: neuronal

communication through neuronal coherence. Trends Cogn Sci2005;9:474–80.

4 Engel AK, Singer W. Temporal binding and the neuralcorrelates of sensory awareness. Trends Cogn Sci 2001;5:16–25.

5 Fernando H, Lopes da Silva. MEG: an introduction tomethods. eds: Hansen, Kringelback & Salmelin. USA: OUP,2010:1–23, figure 1.3 from p6.

6 Buzsáki G, Draguhn A. Neuronal oscillations in corticalnetworks. Science 2004;304:1926–9.

7 Kominis IK, Kornack TW, Allred JC, et al. A subfemtoteslamultichannel atomic magnetometer. Nature 2003;422:596–9.

8 Gross J, Baillet S, Barnes GR, et al. Good practice for conductingand reporting MEG research. Neuroimage 2013;65:349–63.

9 Friston K, Harrison L, Daunizeau J, et al. Multiple sparsepriors for the M/EEG inverse problem. Neuroimage2008;39:1104–20.

10 Woolrich M, Hunt L, Groves A, et al. MEG beamformingusing Bayesian PCA for adaptive data covariance matrixregularization. Neuroimage 2011;57:1466–79.

11 Salmelin R, Service E, Kiesilä P, et al. Impaired visual wordprocessing in dyslexia revealed with magnetoencephalography.Ann Neurol 1996;40:157–62.

12 Hipp JF, Hawellek DJ, Corbetta M, et al. Large-scale corticalcorrelation structure of spontaneous oscillatory activity. NatNeurosci 2012;15:884–90.

13 Stufflebeam SM. Clinical magnetoencephalography forneurosurgery. Neurosurg Clin N Am 2011;22:153–67, vii–viii.

14 Brookes MJ, Woolrich M, Luckhoo H, et al. Investigating theelectrophysiological basis of resting state networks usingmagnetoencephalography. Proc Natl Acad Sci USA2011;108:16783–8.

15 Luckhoo H, Hale JRJRR, Stokes MGMG, et al. Inferringtask-related networks using independent component analysis inmagnetoencephalography. Neuroimage 2012;62:530–41.

16 Smith SM, Fox PT, Miller KL, et al. Correspondence of thebrain’s functional architecture during activation and rest. ProcNatl Acad Sci USA 2009;106:13040–5.

17 Stam CJ. Use of magnetoencephalography (MEG) to studyfunctional brain networks in neurodegenerative disorders.J Neurol Sci 2010;289:128–34.

18 Singh KD. Which ‘neural activity’ do you mean? fMRI, MEG,oscillations and neurotransmitters. Neuroimage 2012;62:1121–30.

19 Fernández A, Maestú F, Amo C, et al. Focal temporoparietalslow activity in Alzheimer’s disease revealed bymagnetoencephalography. Biol Psychiatry 2002;52:764–70.

20 Osipova D, Ahveninen J, Jensen O, et al. Altered generation ofspontaneous oscillations in Alzheimer’s disease. Neuroimage2005;27:835–41.

21 Maestu F, Arrazola J, Fernandez A, et al. Do cognitive patternsof brain magnetic activity correlate with hippocampal atrophyin Alzheimer’s disease? J Neurol Neurosurg Psychiatry2003;74:208–12.

22 Pekkonen E, Ahveninen J, Virtanen J, et al. Parkinson’s diseaseselectively impairs preattentive auditory processing: an MEGstudy. Neuroreport 1998;9:2949–52.

23 Osipova D, Pekkonen E, Ahveninen J. Enhanced magneticauditory steady-state response in early Alzheimer’s disease. ClinNeurophysiol 2006;117:1990–5.

24 Pekkonen E, Osipova D, Laaksovirta H. Magnetoencephalographicevidence of abnormal auditory processing in amyotrophic lateralsclerosis with bulbar signs. Clin Neurophysiol Off J Int Fed ClinNeurophysiol 2004;115:309–15.

25 Stoffers D, Bosboom JLW, Deijen JB, et al. Slowing ofoscillatory brain activity is a stable characteristic of Parkinson’sdisease without dementia. Brain 2007;130:1847–60.

26 Litvak V, Jha A, Eusebio A, et al. Resting oscillatorycortico-subthalamic connectivity in patients with Parkinson’sdisease. Brain 2011;134:359–74.

27 Beckmann CF, DeLuca M, Devlin JT, et al. Investigations intoresting-state connectivity using independent component analysis.Philos Trans R Soc Lond B Biol Sci 2005;360:1001–13.

28 Stam CJ, Jones BF, Manshanden I, et al.Magnetoencephalographic evaluation of resting-state functionalconnectivity in Alzheimer’s disease. Neuroimage2006;32:1335–44.



http://creativecommons.org/licenses/by/3.0/

http://creativecommons.org/licenses/by/3.0/

29 De Haan W, van der Flier WM, Koene T, et al. Disruptedmodular brain dynamics reflect cognitive dysfunction inAlzheimer’s disease. Neuroimage 2012;59:3085–93.

30 Tijms BM, Wink AM, de Haan W, et al. Alzheimer’s disease:connecting findings from graph theoretical studies of brainnetworks. Neurobiol Aging 2013;34:2023–36.

31 Honey CJ, Sporns O, Cammoun L, et al. Predicting humanresting-state functional connectivity from structuralconnectivity. Proc Natl Acad Sci USA 2009;106:2035–40.

32 Douaud G, Filippini N, Knight S, et al. Integration ofstructural and functional magnetic resonance imaging inamyotrophic lateral sclerosis. Brain 2011;134:3470–9.

33 Franciotti R, Iacono D, Della Penna S, et al. Cortical rhythmsreactivity in AD, LBD and normal subjects: a quantitative MEGstudy. Neurobiol Aging 2006;27:1100–9.

34 Babiloni C, Cassetta E, Chiovenda P, et al. Alpha rhythms inmild dements during visual delayed choice reaction time tasks:a MEG study. Brain Res Bull 2005;65:457–70.

35 Hughes LE, Nestor PJ, Hodges JR, et al. Magnetoencephalographyof frontotemporal dementia: spatiotemporally localized changesduring semantic decisions. Brain 2011;134:2513–22.

36 Suntrup S, Teismann I, Bejer J, et al. Evidence for adaptivecortical changes in swallowing in Parkinson’s disease. Brain2013;136:726–38.

37 Teismann IK, Warnecke T, Suntrup S, et al. Cortical processingof swallowing in ALS patients with progressive dysphagia—amagnetoencephalographic study. PLoS ONE 2011;6:e19987.

38 Dziewas R, Teismann IK, Suntrup S, et al. Cortical compensationassociated with dysphagia caused by selective degeneration ofbulbar motor neurons.Hum Brain Mapp 2009;30:1352–60.

39 Maestú F, Yubero R, Moratti S, et al. Brain activity patterns instable and progressive mild cognitive impairment duringworking memory as evidenced by magnetoencephalography.J Clin Neurophysiol 2011;28:202–9.

40 Verdoorn TA, McCarten JR, Arciniegas DB, et al. Evaluationand tracking of Alzheimer’s disease severity using resting-statemagnetoencephalography. J Alzheimers Dis 2011;26(Suppl 3):239–55.

41 Olde Dubbelink KTE, Hillebrand A, Stoffers D, et al.Disrupted brain network topology in Parkinson’s disease:a longitudinal magnetoencephalography study. Brain 2014;137(Pt 1):197–207.

42 Olde Dubbelink KTE, Hillebrand A, Twisk JWR, et al. Predictingdementia in Parkinson disease by combining neurophysiologicand cognitive markers. Neurology 2013;82:263–70.

43 Eisen A, Turner MR. Does variation in neurodegenerativedisease susceptibility and phenotype reflect cerebral differencesat the network level? Amyotroph Lateral Scler FrontotemporalDegener 2013;14:1–7.



Magnetoencephalography

Healthcare

Transcript of Magnetoencephalography